8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
1/12
Eects of fragmentation of the Atlantic forest on mammal
communities in south-eastern Brazil
Adriano G. Chiarello*
Wildlife Research Group, Department of Anatomy, University of Cambridge, UK
Received 1 October 1997; received in revised form 19 August 1998; accepted 21 August 1998
Abstract
Six Atlantic forest reserves, two large (c. 20,000 ha each), two medium-sized (c. 2,000 ha each) and two small (c. 200 ha each),located in northern Esprito Santo, south-eastern Brazil were censused for mammals from October 1994 to April 1996. Diurnal and
nocturnal line-transect sampling was used for censusing mammals>1 kg body weight, and the relative abundance of species in the
six fragments was compared. The number of mammal species recorded in the reserves was strongly related to the forest area, the
richest community being recorded in the two large reserves and the poorest in the two small reserves. The large reserves had a
structurally more complex community, with top predators, large, terrestrial frugivores and large myrmecophages. The mammal
community of the small and the medium-sized reserves was impoverished and less complex. Frugivores were numerically pre-
dominant in both large and medium-sized reserves, whereas herbivores dominated the mammal community of the small reserves,
mainly through the absence of agoutis Dasyprocta leporina and the high density of maned sloth Bradypus torquatus in the two small
reserves. The lack of predators and the proliferation of secondary vegetation and lianas throughout the small reserves are probably
the main causes for the success of arboreal folivores there. Illegal hunting was found to reduce the encounter rates of mostly large,
terrestrial frugivores such as agoutis, pacas Agouti paca, peccaries Pecari tajacu and Tayassu pecari and deer Mazama spp., and is
contributing to keep the population of the surviving species low. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Atlantic forest; Brazil; Forest fragmentation; Mammal community
1. Introduction
When a formerly continuous forest is isolated, the
number of species will shift from its original equili-
brium, mainly because of the eects of area reduction
and distance to continuous forest or between forest
patches. With time, the diversity will decline, eventually
reaching a new, less diverse steady state (MacArthur
and Wilson, 1967; Harris, 1984). The number of species
a habitat can be expected to hold after a period of iso-
lation is strongly area-dependent. The larger the forest
island the higher the original number of species included
and the lower is the rate of subsequent extinctions
(Terborgh and Winter, 1980). Willis (1979), for exam-
ple, comparing the bird community of three Atlantic
forest fragments of diering sizes (1400, 250 and 21 ha)
found 202, 146 and 93 bird species, respectively. Similar
results were found for the community of small mam-
mals in temperate forests (Matthiae and Stearns, 1981)
and for arboreal marsupials in forest isolates of tropical
Australia (Laurance, 1990). There are some conicting
results, however, and the relationship between mamma-
lian species richness and forest size is not always clear-
cut (Mathiae and Stearns, 1981). Generally, the dis-
appearance of species ts a pattern of early loss of large
specialised species, a pattern known as ``ecological
truncation'', probably mainly due to the fact that most
such species occur at very low densities, require large
areas, or both (Wilson and Willis, 1975). The dis-
appearance of ant-following birds of Barro Colorado
Island is a clear example of this phenomenon.
In this paper I present a detailed analysis of the com-
munity of large and medium-sized mammals currently
present in six remnants of the lowland Atlantic forest.
Originally this forest extended almost continuously
from the state of Rio Grande do Norte, in north-east
Brazil, to the Rio Grande do Sul, the southernmost
state of the country, stretching over an area of 1,200,000
km2 or ca. 12% of the country (Brown and Brown,
1992). As this region coincides with the highest density
of urban settlements, it has been cleared, disturbed, or
0006-3207/99/$see front matter# 1999 Elsevier Science Ltd. All rights reserved.
P I I : S 0 0 0 6 - 3 2 0 7 ( 9 8 ) 0 0 1 30 - X
Biological Conservation 89 (1999) 7182
* Current address: Museu de Biologia Prof. Mello Leitao, SantaTeresa, ES, 29650-000, Brazil. E-mail: [email protected]
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
2/12
burnt almost continuously since the arrival of the rst
Europeans in the early 1500s (Fonseca, 1985; Dean,
1995). Currently, 43% of the Brazilian population is
concentrated in this region (Fonseca, 1985), and conse-
quently, only about 512% of the original forest
remains (Brown and Brown, 1992) as isolated forest
remnants, scattered throughout a landscape dominatedmainly by pasturelands and agricultural elds.
The main objectives of this paper were: (1) to identify
the mammal species richness and the relative abundance
of individuals surviving in each reserve, and (2) to ana-
lyse changes in the structure of the mammal community
contrasting the relative abundances of orders and diet-
ary categories between reserves of diering sizes. Given
that community studies with large mammals of the
Atlantic forest are lacking, the results reported here are
compared mainly to both small and large mammals
studied elsewhere in the tropics.
2. Methods
2.1. Study areas
Data were collected in six Atlantic Forest reserves,
varying in size from 210 to 24,250 ha, located in north-
ern Esprito Santo state in south-eastern Brazil (1812H
1948HS; 3950H4015HW; Fig. 1). This region has been
disturbed since the early 1500s, but forest destruction
increased greatly during the early 20th century (Aguirre,
Fig. 1. Detail of Northern Esprito Santo state, in south-eastern Brazil, showing the patches of native forests (dark areas) remaining in 1995, urban
areas (open circles), and the study sites (16) (based on satellite images from SOS Mata Atla ntica and INPE, 1997). Large reserves: LFR (1) and
SBR (2); medium-sized reserves: CVBR (3) and CGBR (4), and small reserves: M7/317 (5) and Putiri (6).
72 A.G. Chiarello/Biological Conservation 89 (1999) 7182
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
3/12
1951). The principal cause of deforestation was logging
(Heinsdijk et al., 1965), which is still taking place in the
region, as satellite images indicated that 5.5% of the
state's area was cleared of forests between 1990 and
1995 (SOS Mata Atla ntica et al., 1998). The most recent
estimate available (1995) revealed that ` 97 of the
state area is still covered with native forests, and themajority of the remnants are small and isolated forest
patches of ` 1000 ha of area (SOS Mata Atla ntica et
al., 1998; Fig. 1).
Three of the study areas are privately owned reserves
(LFR, M7/317 and Putiri) and the other three (SBR,
CVBR and CGBR) are biological reserves administered
by the Brazilian Environmental Agency (IBAMA). All
reserves exhibit signs of past human disturbance,
including illegal hunting, intrusion of res from adja-
cent pasturelands, and logging, but large tracts of pre-
dominantly primary forest are still found in the two
larger reserves (Table 1). The extreme north of LFR is
contiguous with the other large reserve (SBR) but thelatter is crossed in its eastern part by a highway (Fig. 1).
The remaining study areas are isolated from each other
and from other forest fragments. The landscape matrix
in which the large and medium-sized reserves are
immersed is composed predominantly of pasturelands
and agricultural elds (mainly coee and sugar cane),
whereas extensive plantations of Eucalyptus spp. for the
production of cellulose, surround the two small reserves
(M7/315 and Putiri).
The reserves have similar climatic, edaphic, topo-
graphic and physiognomic conditions, and, therefore,
their fauna and ora were assumed to be composedoriginally of similar sets of species. The predominant
vegetation in all reserves is the tropical rain forest of the
Tertiary tablelands (``Mata de Tabuleiros''), which is a
semi-deciduous, mesophytic forest formation of the
Atlantic forest domain (Rizzini, 1963). The soils of this
region are predominantly acidic Tertiary sediments
(Hiensdijk et al., 1965) with poor to average fertility
(Moraes, 1974), and the topography is at with altitudes
of 3090 m (Moraes, 1974). The predominant climate is
the Aw of Koppen: hot and humid with a dry season
during autumnwinter (AprilSeptember) and a wet
season during springsummer (OctoberMarch). The
climate is similar for the six study areas, with averagetemperatures of 2325C and annual rainfall of 1050
1420 mm (unpublished data from weather stations of
Companhia Vale do Rio Doce SA and Aracruz Celulose
SA).
2.2. Data collection
The sampling protocol in all study areas consisted of
line transect sampling sensu Buckland et al. (1993). All
mammal species seen during censusing, both during the
day and at night, were included in the analysis, except
bats and small mammals such as rodents and marsu-
pials, which require specic sampling methods beyondthe scope of this study. The only exceptions were the
common opossum Didelphis aurita and the bare-tailed
woolly opossum Caluromys philander, seen frequently
during transect sampling.
Straight trails of 15002000 m of length and 1.5 m
wide were cut in each study site in locations selected to
include all forest types present in the reserve or frag-
ment in question. Non-forested areas such a native open
elds and marsh areas were not sampled, but some parts
of the forest edges were used. Trails were allowed to
``rest'' for a month before the beginning of sampling
and were not censused on consecutive sampling days.Diurnal censuses started before sunrise and were termi-
nated after 34 h. The nocturnal censuses started after
sunset and were nished, on average, after 34 h. Eort
was made to keep the walking speed as constant as
possible at ca. 1 km/h. About 30 censuses were carried
out in each reserve, totalling 175 censuses in the six
Table 1
Area, approximate time elapsed since isolation, disturbance level, predominant forest type, and hunting pressure of the six reserves included in this
study
Study area Area (ha)
Years since
isolation
Disturbance
levelaPredominant
forest typebHunting
pressure
Large reserves
Linhares Forest Reserve (LFR) 21,800 30 Light Primary Low
Sooretama Biological Reserve (SBR) 24,250 30 Light Primary High
Medium-sized reserves
Co rrego do Veado Biological Reserve (CVBR) 2400 30 High Secondary Moderate
Co rrego Grande Biological Reserve (CGBR) 1504 1015 Moderate Secondary Moderate
Small reserves
M7/317 260 25 Moderate Secondary Moderate
Putiri 210 23 Moderate Secondary Moderate
a Forest disturbance caused by accidental intrusion of res, clearings, and selective logging in the past.b Type of forest currently covering more than half of the reserve's area.
A.G. Chiarello/Biological Conservation 89 (1999) 7182 73
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
4/12
study areas, 465.9 h of sampling and a total transect
length of 458.8 km (Table 2). Unfortunately, time and
logistic limitations prevented an equal amount of eort
allocated to nocturnal and diurnal sampling, but the
time allocated to nocturnal and diurnal sampling wassimilar between reserves (Table 2).
Here the term ``visual encounter'' refers to those
occasions in which data such as species, group size and
composition (if social) could be collected for the speci-
men in question. Binoculars (8 42) were used during
all observations. For the nocturnal censuses, a 55-watt
hand-held spotlight was used. Data on calls, footprints,
faeces, and carcasses found in the reserves during the
study were also recorded as additional evidence of a
species' presence there. Interviews were conducted with
some local woodsmen, former hunters and the sta of
all reserves, in order to collect additional informationon mammal species not seen during the study.
2.3. Data analysis
Encounter rates were used to compare the relative
abundance of mammal species between reserves (Janson
and Emmons, 1990). This method was chosen because,
for most species, the number of encounters was not
large enough to estimate their true densities without
incurring serious bias. Encounter rates were calculated
taking into account the species' period of activity. For
all species, the total transect length was calculated as the
summation of the lengths of all individual trails cen-
sused. For those species active both during the day and
at night, for example, armadillos, tamanduas and pecc-
aries, the total transect length included both diurnal and
nocturnal sampling. Encounter rates were calculated as
the total number of encounters/10 km for all mammal
species seen during transect sampling, and these data
were used in a Hierarchical Cluster Analysis in order to
arrange reserves into groups (Dillon and Goldstein,
1984). Following Kent and Coker (1996), the Ward's
minimum variance was chosen as the preferred method
of similarity analysis and the squared Euclidean dis-
tance was used as a measure of dissimilarity. Data on
encounter rates were grouped by orders following the
classication of Wilson and Reeder (1993) and by diet-
ary categories following the scheme originally proposed
by Eisenberg (1981) and latter modied by Robinson
and Redford (1986).
Multiple regression analysis was used to assess theinuence that habitat variables, such as fragment size,
time since isolation, disturbance level and hunting pres-
sure, have on species richness and abundance (encoun-
ter rate). The two categorical variables, disturbance
level and hunting pressure, were coded, from lowest to
highest, as follows: disturbance level=1 (LFR and
SBR), 2 (CGBR, M7/317 and Putiri) and 3 (CVBR);
hunting pressure=1(LFR); 2 (CVBR, CGBR, M7/317
and Putiri), and 3 (SBR). The disturbance degree of
each reserve was ranked taking into account the pro-
portion of area covered by secondary vegetation,
including cleared and burnt areas, and by signs of past
logging activities (Chiarello, 1997). Hunting pressurewas assessed by the number of occasions shots were
heard during censuses, by the number of hunter trails
found, and by the frequency of encounters with hunters
or their dogs during censusing (Chiarello, 1997). After
several experimental runs, a stepwise method was cho-
sen in which these four habitat variables were included
or excluded from the regression equation after passing a
signicant test, with a signicance of 0.10 for inclusion
(p in) and 0.20 for exclusion (p out). All statistical ana-
lyses were carried out in IBM-compatible computers
with the SPSS software package (Norusis, 1993), and all
probabilities reported here are 2-tailed.
3. Results
3.1. Species richness
Table 3 lists the 37 mammal species recorded in the
area of the six reserves and fragments during the present
study. The number of species not recorded in the two
small reserves (M7/317 and Putiri) was about twice the
gure observed for the medium-sized reserves (CVBR
and CGBR), but only one species found in smaller
reserves, the maned sloth Bradypus torquatus was not
recorded in the two large reserves. The number of spe-
cies in each reserve showed a positive relationship with
forest area, i.e. the larger the reserve's area the higher
the number of species recorded in it (multiple regression
analysis; R2 0X960; F 96X801; p 0X006). Fragment
size was the only independent variable selected by the
stepwise method in this analysis (regression coecient
or Beta 0X980; p ` 0X001); time since isolation, dis-
turbance level and hunting pressure were excluded from
the nal equation. Conversely, the number of species
extinct or otherwise not recorded in the reserves was
Table 2
Number and length (km) of diurnal and nocturnal censuses of mam-
mals in the six study areas
Diurnal Nocturnal Total
Area No. Length No. Length No. Length
LFR 21 64.9 9 15.2 30 80.1SBR 21 66.1 11 17.5 32 83.6
CVBR 19 66.4 13 18.9 32 85.3
CGBR 18 66.9 11 21.0 29 87.9
M7/317 17 45.6 10 17.3 27 62.9
Putiri 16 48.9 9 10.0 25 59.0
Total 112 358.9 63 99.9 175 458.8
74 A.G. Chiarello/Biological Conservation 89 (1999) 7182
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
5/12
negatively related to reserve area (R2 0X961;
F 98X996; p ` 0X001), and again fragment size was
included in the nal regression with a signicant coe-
cient (Beta 0X980; p ` 0X001).
The two large reserves (LFR and SBR) presented the
richest fauna and virtually the same composition of
species, which included three species found exclusively
there: the giant armadillo Priodontes maximus, the
jaguar Panthera onca and the gray brocket deer
Mazama gouazoupira. The presence of other large-bod-
ied species such as tapirs Tapirus terrestris and two
peccaries, Pecari tajacu and Tayassu pecari, were also
conrmed there. These two large reserves had the most
diverse fauna of predators, composed by a total of 10
species, eight of which potentially capable of taking
mammals 51 kg of body weight, namely two big cats,
the jaguar and puma Puma concolor, the ocelot Leo-
pardus pardalis, three small cats Herpailurus yaguarondi,
Table 3
Mammal species seen, heard, reported by local people, or whose footprints or faeces where found in the six reserves during the present study c
Large Medium-sized Small
Common namea Scientic nameb LFR SBR CVBR CGBR M7/317 Putiri
1 Bare-tailed wooly opossum Caluromys philander ? v ? v
2 Southeastern common opossum Didelphis aurita v v v r r v3 Maned sloth Bradypus torquatus r v v
4 Brown-throated three-toed sloth Bradypus variegatus v r r
5 Southern naked-tailed armadillo Cabassous unicinctus v ? r ?
6 Nine-banded long-nosed armadillo Dasypus novemcinctus v v v r v v
7 Giant armadillo Priodontes maximus r r
8 Southern tamandua Tamandua tetradactyla r r c v v ?
9 Giant anteater Myrmecophaga tridactyla ? ? ? ?
10 Tufted-ear marmoset Callithrix georoyi v v v v r v
11 Masked titi monkey Callicebus personatus v v a v v
12 Brown capuchin monkey Cebus apella v v v v v v
13 Brown howler monkey Alouatta fusca v a,fa ? v
14 Crab-eating fox Cerdocyon thous v fa v v v v
15 Jaguarundi Herpailurus yaguarondi v r r r v ?
16 Ocelot Leopardus pardalis r r ? ?
17 Oncilla Leopardus tigrinus r v r ? r ?18 Margay Leopardus wiedii r ? r v r ?
19 Puma Puma concolor r,fo r,fa fa,r
20 Jaguar Panthera onca v r
21 Tayra Eira barbara r r r v v ?
22 Grison Galictis vittata r r r ? ? ?
23 South American coati Nasua nasua v v v v v v
24 Crab-eating raccoon Procyon cancrivorus v fo r fo fo ?
25 Kinkajou Potos avus v r v v
26 Brazilian tapir Tapirus terrestris v r v fo,a
27 Collared peccary Pecari tajacu v r v
28 White-lipped peccary Tayassu pecari v r v
29 Gray brocket deer Mazama gouazoupira v v
30 Red brocket deer Mazama americana r r r ? v v
31 Guianan squirrel Sciurus aestuans v v v v v v
32 Bahia hairy dwarf porcupine Sphiggurus insidiosus r v v ? ? v33 Capybara Hydrochaeris hydrochaeris fo,fa r ? ? ? ?
34 Red-humped agouti Dasyprocta leporina v v v v
35 Paca Agouti paca v fo v v r r
36 Bristle-spined porcupine Chaetomys subspinosus ? ? ? ? r v
37 Tapiti Sylvilagus brasiliensis v v v r v ?
Presence conrmed (1) 33 32 25 21 20 13
Presence probable (2) 3 4 5 10 3 9
Total surviving (1) + (2) 36 36 30 31 23 22
Extinct or absent 1 1 7 6 14 15
a Common names from Emmons and Feer (1997).b Scientic names from Wilson and Reeder (1993).c Abbreviations: (a) auditory evidence (calls or noise when ushed); (c) carcass found; (fa) faeces found; (fo) footprints; (r) reported by other
researchers or local people; (v) visual encounter; (?) presence probable but not conrmed during study; () extinct or absent (i.e. never cited for thelocation).
A.G. Chiarello/Biological Conservation 89 (1999) 7182 75
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
6/12
Leopardus wiedii, L. tigrinus, the tayra Eira barbara and
the crab-eating fox Cerdocyon thous. The two medium-
sized reserves (CVBR and CGBR) presented a slightly
less diverse fauna but interesting dierences were found
between them. While the jaguar was not recorded in
either the medium-sized reserves, evidence of the
presence of the other large cat (puma), was found in theCGBR. The tapir was recorded in both reserves, but
only the collared peccary P. tajacu was observed in
CGBR and the white-lipped peccary T. pecari in the
CVBR. The most striking dierence between study
areas was the total absence of terrestrial, large-bodied
species such as tapirs, peccaries, giant armadillos, and
predators such as jaguars and pumas in the two small
fragments (M7/317 and Putiri). Another notableabsence in these fragments was the agouti Dasyprocta
Table 4
Number of visual encounters with mammals/10 km of transect sampling for the six study areas
Large Medium-sized Small
Species/ordera Dietary categoryc Habitd LFR SBR CVBR CGBR M7/317 Putiri Mean
Didelphimorphia
1 Caluromys philander Fr/om Arb 0 0.57 0 0.95 0 0 0.25
2 Didelphis aurita Fr/om Sca 0.66 1.14 2.65 0 0 1.00 0.91
Xenarthra3 Bradypus torquatus He/br Arb 0 0 0 0 5.70 4.70 1.73
4 Dasypus novemcinctus In/om Ter 0.12 0.12 0.23 0 0.16 0.17 0.13
5 Tamandua tetradactyla Myr Sca 0 0 0 0.34 0.16 0 0.08
Primates
6 Callithrix georoyib Fr/om Arb 2.16 1.81 1.05 0.15 0 0.20 0.90
7 Callicebus personatusb Fr/he Arb 1.23 1.66 0 0 0.22 1.02 0.69
8 Cebus apellab Fr/om Arb 2.47 1.51 1.05 0.60 2.19 1.23 1.51
9 Alouatta fuscab Fr/he Arb 0.15 0 0 0 0.22 0 0.06
Carnivora
10 Cerdocyon thous Fr/om Ter 0 0 0 0.48 0 0 0.08
11 Leopardus tigrinus Car Ter 0 0.12 0 0 0 0 0.02
12 Herpailurus yaguarondi Car Ter 0.12 0 0 0 0 0 0.02
13 Eira barbara Fr/om Sca 0 0 0 0.15 0.22 0 0.0614 Potos avus Fr/om Arb 0 0 1.06 2.38 0 0 0.57
15 Nasua nasuab Fr/om Sca 0.15 0.30 0.60 0.60 0.22 0.20 0.35
Perissodactyla
16 Tapirus terrestris Fr/he Ter 0.12 0 0.23 0 0 0 0.06
Artiodactyla
17 Pecari tajacub Fr/he Ter 0 0 0 1.02 0 0 0.17
18 Tayassu pecarib Fr/he Ter 0 0 0.12 0 0 0 0.02
19 Mazama spp. Fr/he Ter 2.50 0.60 0 0 0.95 0.34 0.73
Rodentia
20 Sciurus aestuans Fr/gr Arb 7.55 6.80 10.54 8.22 2.85 3.68 6.61
21 Sphiggurus insidiosus Fr/gr Arb 0 0 0.23 0 0 0.17 0.07
22 Dasyprocta leporina Fr/gr Ter 6.01 1.97 3.46 0.30 0 0 1.9623 Agouti paca Fr/gr Ter 1.32 0 1.06 0.48 0 0 0.48
24 Chaetomys subspinosus He/br Arb 0 0 0 0 0 0.34 0.06
Lagomorpha
25 Sylvilagus brasiliensis He/gz Ter 1.98 0.57 0.53 0 1.16 0 0.71
Mean 19.6 12.4 15.0 10.0 10.2 10.3 13.1
a Following Wilson and Reeder (1993).b Species whose encounter rates are given in groups/10 km.c Dietary categories from Eisenberg (1981) and Robinson and Redford (1986): He/gz (herbivoregrazer), He/br (herbivorebrowser), Fr/he
(frugivoreherbivore), Fr/gr (frugivoregranivore), Fr/om (frugivoreomnivore), In/om (insectivoreomnivore), Myr (myrmecophage), and Car
(carnivore).d Habits: Ter (terrestrial), Sca (scansorial), Arb (arboreal).
76 A.G. Chiarello/Biological Conservation 89 (1999) 7182
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
7/12
leporina. The mammalian faunas of the two small frag-
ments (M7/317 and Putiri), although depleted in terms
of species number, were very similar.
3.2. Species abundance
The rate of encounter/10 km of census walked is
presented in Table 4 for the 25 mammal species seenduring transect sampling, representing 68% of the
mammal fauna known to occur in the six study areas.
This species list resulted from 602 encounters with
mammals (509 encounters if individuals ushed during
censuses are not taken into account). A signicant rela-
tionship (Multiple regression analysis; R2 0X925;
F 18X512; p 0X021) was found between encounter
rate, the dependent variable, and two independent vari-
ables: fragment size (Beta 0X708; p 0X045) and
hunting pressure (Beta 0X610; p 0X066). The step-
wise method did not include time since isolation and
disturbance level in the nal regression equation ofthis analysis.
The average gure for the six reserves and for all
mammal species was 13.1 encounters/10 km, but sig-
nicant dierences were observed between study areas.
On average, mammals were about twice as abundant in
one large reserve (LFR), as in the two small reserves
(M7/317 and Putiri) and the medium-sized CGBR. The
other medium reserve (CVBR) presented the second
highest abundance of mammals, and the third gure
was recorded for the large SBR. The Guianan squirrel
Sciurus aestuans and the agouti were, respectively, the
rst and second most commonly seen species in LFR,
SBR and CVBR. On the other hand, the maned sloth
was the most abundant mammal encountered in the two
small fragments.
3.2.1. Encounter rates
The two small fragments (M7/317 and Putiri) were
very similar in terms of encounter rates, with mammal
species forming an external cluster unlike the other four
study areas (Fig. 2). The other four reserves formed a
separate group of which LFR was the most divergent.
Considering data from all six reserves together, the top
three orders in relative abundance were rodents, pri-
mates and xenarthrans, respectively (Fig. 3), but two
distinct patterns emerged among the study areas. First,
similar-sized reserves had similar faunas. Secondly,
large and medium-sized reserves had similar abun-
dances of orders, with a clear dominance of rodents
(Fig. 3a and b), unlike the two small reserves (Fig. 3c).
Large and medium-sized reserves diered slightly, how-
ever, in the abundance of primates, which were more
abundant in the two largest areas (12 4X788; dXfX 1;
p ` 0X05; Fig. 3a). In the small reserves, xenarthrans
were as abundant as rodents (12 0X827; dXfX 1;
p b 0X50; Fig. 3c).
3.2.2. Dietary categories
While frugivores were the dominant category in the
four larger reserves (Fig. 4a and b), herbivores had the
highest abundance in the two small-sized reserves (Fig. 4c).
Fig. 2. Dendrogram produced by the Hierarchical Cluster Analysis ofencounter rates with 25 mammal species during transect sampling in
the six study sites. The scale (025) is a measure of distance between
clusters. Large reserves (LFR and SBR), medium-sized reserves
(CVBR and CGBR), small reserves (M7/317 and Putiri).
Fig. 3. Number of encounter/10 km of transect sampling for mam-
malian orders in the six study sites. (a) Large reserves (LFR and SBR);
(b) medium-sized reserves (CVBR and CGBR), and (c) small reserves
(M7/317 and Putiri). Orders: Did (Didelphimorphia), Xen (Xenar-
thra), Pri (Primates), Car (Carnivora), Per (Perissodactyla), Art
(Artiodactyla), Rod (Rodentia), and Lag (Lagomorpha).
A.G. Chiarello/Biological Conservation 89 (1999) 7182 77
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
8/12
This dierence was due to the complete absence of
agoutis and to the lower abundance of squirrels in both
M7/317 and Putiri (Table 4). Within herbivores, herbi-
vore-browsers, represented by sloths Bradypus spp. and
by the bristle-spined porcupine Chaetomys subspinosus,
was the category that exhibited major dierences in
density among study sites.
3.2.3. Habit and hunting
Major dierences between study areas occurred in
density of terrestrial and, to a lesser extent, scansorial
species (Fig. 5). Terrestrial species were more abundant
in large and medium-sized reserves (LFR, CVBR and
SBR), and the lowest densities were found in the small
fragments, but notably in Putiri, where only one terres-
trial mammal was encountered, on average, in each 20
km of censusing (0.5 encounter/10 km). The two large
reserves had contrasting abundances of terrestrial
mammals as a much lower density was found in SBR
than in LFR (Fig. 5). Since these two reserves had
contrasting hunting pressures, the Chi-square test was
used to check if hunting could account for this dier-
ence, comparing the encounter rates of hunted and non-hunted species between these two reserves. For hunted
species (tapirs, peccaries, brocket deer, agoutis, pacas
Agouti paca, tapitis Sylvilagus brasiliensis and arma-
dillos Dasypus novemcinctus) the encounter rate was
signicantly higher in LFR (12.05 encounter/10 km)
than in SBR (3.26 encounter/10 km) (12 5X048;
dXfX 1; p ` 0X025), but no statistical dierence was
found for non-hunted species (all other species;
vp 14X49 encounter/10 km; f 13X91 encoun-
ter/10 km; 12 0X012; dXfX 1; p b 0X90).
4. Discussion
4.1. Species richness
The number of mammal species recorded in the study
areas was highest in the large reserves, intermediate in
the medium-sized reserves and lowest in the two small
fragments, as classically predicted by the island biogeo-
graphy theory (MacArthur and Wilson, 1967). Extinc-
tion becomes the predominant phenomenon that
accounts for this process of species loss in forest frag-
ments because, being isolated by non-forest vegetation,
they lack the sources of potential new migrants. In
agreement with the nested subset hypothesis of Patter-
son (1987), the mammalian faunas recorded here for the
small and medium-sized fragments were, indeed, subsets
of those in larger reserves. This was because extinction-
prone species were progressively lost from fragments
over time, with smaller fragments exhibiting the greatest
losses. Several studies on forest fragmentation have
demonstrated that larger fragments support higher spe-
cies richness than smaller fragments (Matthiae and
Stearns, 1981; Fonseca and Robinson, 1990; Laurance,
1990). A number of factors may contribute to this, the
Fig. 4. Number of encounter/10 km of transect sampling for dietarycategories in the six study sites. (a) Large reserves (LFR and SBR); (b)
medium-sized reserves (CVBR and CGBR), and (c) small reserves
(M7/317 and Putiri). Dietary categories: He/gz (herbivoregrazer), He/
br (herbivorebrowser), Fr/he (frugivoreherbivore), Fr/gr (frugivore
granivore), Fr/om (frugivoreomnivore), In/om (insectivoreomni-
vore), Myr (myrmecophage), and Car (carnivore).
Fig. 5. Number of encounters/10 km of transect sampling for arbor-
eal, scansorial and terrestrial species in the six study areas.
78 A.G. Chiarello/Biological Conservation 89 (1999) 7182
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
9/12
most obvious one being the fact that, as the forest area
is fragmented and reduced, some species, especially
those with the largest spatial requirements, can not nd
sucient food or other vital resources to survive in the
long term in a smaller area (McNab, 1963; Redford and
Robinson, 1991). Thus large cats, tapirs, peccaries,
giant armadillos and the giant anteater Myrmecophagatridactyla were not recorded in the smaller fragments.
For these species, forest areas of c. 200 ha are probably
not large enough, while the jaguar and the giant arma-
dillo, for example, were not recorded even in the med-
ium-sized reserves of 2000 ha, where only one peccary
species and only one deer species were recorded. The
only species not recorded in the large reserve was the
maned sloth, but this is almost certainly not a case of
extinction, since the species has never been reported to
occur there or in other forest patches of northern
Esprito Santo, where, instead, the three-toed sloth B.
variegatus is believed to occur (Oliver and Santos, 1991;
pers. obs.).In a fragmented forest, top predators, such as jaguars
and pumas, are clearly among the most vulnerable spe-
cies as they occur at low densities and occupy very large
home ranges. Apart from the two large reserves, the
presence of puma was conrmed only in the mid-sized
CGBR, where a scat was found, although it is probably
just a temporary visitor there, because fresh scats and
footprints were not found there during the study. Simi-
larly, the jaguars and pumas that once inhabited the
1500 ha Barro Colorado Island in the Panama Canal
Zone, disappeared from there after that reserve became
an island isolated by the rising waters of Gatun Lakewhen a dam was built to form the Panama Chanel
(Glanz, 1990).
Although faunal inventories were not carried out in
the study areas before the advent of fragmentation, the
available evidence suggests that those species not recor-
ded in both medium- and small-sized reserves during the
present study were originally present there. Based on
reports of early naturalists that travelled through the
area in early 1800s (Wied-Neuwied, 1820; Saint-Hilaire,
1833), all the region to the north of present-day Aracruz
was originally covered by an immense, unbroken forest.
At that time, Wied-Neuwied (1820) listed for this region
all six cat species, today found only in the two large
reserves, as well as tapirs, two species of peccaries, two
species of brocket deer and the giant anteater. The con-
temporaneous (1818) observations of Saint-Hilaire
(1833) added to this list the elusive giant armadillo.
Most of those species were relatively common in the
past and were reported by local residents, or by former
woodcutters, to occur in the study areas until the 1960s
or early 1970s. Additionally, the faunal inventories car-
ried out by Ruschi (1978) in the region of the small
fragments, listed all 15 species not recorded there during
the present study.
4.2. Species abundance
The relative abundance of individual mammal species
varied between study areas, but no noticeable increase
in encounter rates or absolute densities was found in
small and medium-sized reserves in comparison with the
large areas. This has been reported for small mammalcommunities in the fragments of Manaus (Malcolm,
1988), but they are not comparable to those studied
here, because they are both near to and not completely
isolated from the continuous forest. One cannot say that
migration of large mammals never takes place between
forest remnants, given that several fragments of native
vegetation still exist in northern Esprito Santo. As
mentioned above, the puma may roam between frag-
ments, and species that can forage in several habitat
types, such as the crab-eating fox, the crab-eating
racoon Procyon cancrivorous and brocket deer, might
easily cross through non-forest areas. But the potential
for migration is certainly reduced there, as the frag-ments are small and widely separated. Moreover, most
of them are probably overhunted and the majority have
been highly degraded by edge eects or anthropogenic
activities such as logging and res. Fragmentation thus
prevents a hunted population from being replenished
through immigration and limits their movements across
the landscape (Robinson, 1996).
Analysis of relative abundances showed that both
large and medium-sized reserves had a marked pre-
dominance of rodents over any other mammalian order,
and frugivores were more abundant there than herbi-
vores. On the other hand, xenarthrans were the mostabundant order in the two small reserves, and herbi-
vores were encountered there more frequently than any
other dietary category, including frugivores. Two main
groups of factors may account for these dierences.
First, frugivores need larger home ranges than herbi-
vores because fruit, compared to leaves or herbs, have a
patchier distribution both in space and time (Milton
and May, 1976), as fruiting trees are generally not dis-
tributed uniformly throughout the forest and seasons
dier enormously in the amount and quality of fruit
sources available to consumers (Gentry and Emmons,
1987). In the Brazilian Amazon, for example, the white-
lipped peccary and the strictly frugivorous bearded saki
Chiropotes satanas were among the rst species to
abandon a small forest fragment shortly after its isola-
tion from the continuous forest (Lovejoy et al., 1986).
Secondly, the forests of M7/317 and Putiri, in com-
parison to the other areas, present an intensied damage
due to a greater exposure to winds and hot-air curents.
The combination of elevated tree mortality and the
consequent proliferation of secondary vegetation,
including lianas and vines, observed in the small frag-
ments (Chiarello, 1997) are certainly detrimental to fru-
givores but, on the other hand, can aid arboreal
A.G. Chiarello/Biological Conservation 89 (1999) 7182 79
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
10/12
folivores, like the maned sloth, the brown howler mon-
key Alouatta fusca and perhaps the bristle-spined por-
cupine. It is known that these three folivores can survive
well in small patches of degraded forests, because of
their ability to include secondary vegetation in the diet
(Chiarello, 1994; Chiarello et al., 1997; Chiarello, 1998).
Apart from the increased availability of food source(leaves), these arboreal folivores were greatly beneted
in the small fragments from the total absence of pre-
dators such as large cats and harpy eagles Harpia harpyja.
It has been suggested that, in the absence of pre-
dators, local species richness would diminish because
strong competitors among prey species would ultimately
dominate the community (Paine, 1966). Fonseca and
Robinson (1990), for example, noticed that some small
forest patches of the Atlantic forest were dominated by
the common opossum, a generalist species that, in the
absence of predators, outcompeted other species of
small mammals. It would be expected, therefore, that
some prey species of the small reserves would achievehigher densities there than in the large areas, which have
a full complement of both avian and mammalian pre-
dators. However, an increase in relative density was
observed in small and medium-sized reserves for only a
very few species and only those not normally taken by
hunters (see below). The only species which showed very
high densities in the small fragments (M7/317 and
Putiri) was the maned-sloth, which has beneted from
the absence of predators and abundance of food (leaves)
there. All other mammal species found in the two frag-
ments exhibited relative densities equal to or, in most
cases, lower, than that recorded for the other areas.The impoverishment of animal communities in small
forest fragments may have important consequences for
the recruitment of tree species that depend directly or
indirectly on them as pollinators, or for seed dispersal
and predation. It is hypothesised, for example, that cer-
tain tree species such as Hymenaea courbaril (Legumi-
nosae) are wholly dependent on agoutis for dispersal
(Hallwachs, 1986), while Dirzo and Miranda (1990)
point out that the extirpation of herbivorous and gran-
ivorous mammals has profound eects on forest regen-
eration. Putz et al. (1990) found that only one or two
tree species dominate the tiny islets (ca. 1 ha) formed 80
years ago in the Panama Canal Zone during the dam-
ming of the River Chagres as they are too small to har-
bour terrestrial species of frugivorous mammals.
4.3. Impact of hunting
Evidence of ``recreational'' or even commercial hunt-
ing were found in all study areas (Chiarello, 1997) but
the hunting pressure seems to be lowest in LFR as this
reserve has an active system of vigilance against poa-
chers (pers. obs.), and is higher in all other areas, parti-
cularly in SBR, which is under the greatest hunting
pressure of all study sites. The faunal comparison
between SBR and the contiguous LFR revealed a much
lower abundance of game species in SBR, notably
pacas, agoutis and brocket deer. If the impact of hunt-
ing is severe in extensive areas of ``pristine'' forests
(Redford, 1992), the consequences for isolated forest
fragments can be disastrous, as hunters have greateraccess to forests in a fragmented landscape (Robinson,
1996) and most fragments lack the sources of potential
new migrants to replenish hunted populations. Strict
frugivores taken by hunters are especially vulnerable in
small reserves. For example, the agouti is among the
most procured prey by hunters in several localities in the
Neotropics (Redford and Robinson, 1987), and this
may explain why it is completely absent from M7/317
and Putiri, as well as from other reserves of similar or
smaller size (pers. obs.).
The absence in these small fragments of tapir, two
species of peccaries, the gray brocket deer and the
agouti would be, theoretically, benecial to the pro-liferation of the two species of terrestrial frugivores
surviving there, that is, the paca and the red brocket
deer M. americana, but these two species were not
encountered there more frequently than in large and
medium-sized reserves, most probably because their
population are being kept low by hunting. Similarly, in
medium-sized reserves, apart from the collared peccary,
only those species not subjected to hunting exhibited
higher densities. The evidence found here suggests,
therefore, that hunting in isolated forest fragments is an
``external'' force that counteracts the inuence of biolo-
gical processes, such as competition and predation(MacArthur et al., 1972).
The results presented here show that isolated Atlantic
forest fragments of 200 ha or less were too small and
disturbed to maintain intact assemblages of mammals.
Smaller fragments not only had an impoverished
assemblage of species, but also, the few mammal species
surviving in them have very small population sizes, most
of which are less then 50 individuals/reserve (Chiarello,
1997). Such small populations are highly susceptible to
stochastic factors, both demographic and environ-
mental, and to genetic deterioration due to inbreeding
and loss of allelic diversity (Frankham, 1998; Franklin
and Frankham, 1998). Consequently, their chances of
survival in the long term are very limited. After 2030
years of isolation, only reserves with ca. 20,000 ha or
more still harbour a practically intact mammalian
fauna, from herbivores to top predators. The main
implication is that, in the long term, large mammals of
the Atlantic forest have good chances of survival only in
a small number of reserves, as the great majority of
forest remnants in this biome are disturbed, isolated,
and have ` 1000 ha of area (SOS Mata Atla ntica et al.,
1998). The loss of an important component of this bio-
diversity hotspot (Mittermeier et al., 1998) is, therefore,
80 A.G. Chiarello/Biological Conservation 89 (1999) 7182
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
11/12
already well underway and will tend to increase in the
next few decades.
Acknowledgements
The author is most grateful to David Chivers for hissupport and advice throughout the project. My study in
Cambridge was made possible by a doctoral grant from
the Brazilian Science Council (grant no. 200273/92-2).
The National Geographic Society (no. 5365-94), World
Fund for Nature-Brazil (no. CBO 123-94), Fauna and
Flora International (no. 94/32/10), Cambridge Overseas
Trust and Girton College contributed funds for the eld
work. Authorisation to carry out eld work and help
with logistics was provided by The Brazilian Environ-
mental Agency, Companhia Vale do Rio Doce SA, and
Aracruz Celulose SA. Clarice Bassi, Daniel Borges,
Rudi Laps, and Susanna Busch, helped with data col-
lection. I also thank Se rgio Mendes and Marlon Zorte afor allowing me and my eld assistants to use the facil-
ities and equipment of Museu de Biologia Mello Leitao
as well as for the help with local bibliography. Brian
Davis and two anonymous reviewers made helpful
comments on early drafts.
References
Aguirre, A., 1951. Sooretama, Estudo sobre o Parque de Reserva,Refu gio e Criac ao de Animais Silvestres, ``Sooretama'', no Munic-pio de Linhares, Estado do Esprito Santo. Ministe rio da Agri-cultura, Servic o de Informac ao Agrcola, Rio de Janeiro.
Brown Jr, K.S., Brown, G.G., 1992. Habitat alteration and speciesloss in Brazilian forests. In: Whitmore, T.C., Sayer, J.A. (Eds.),Tropical Deforestation and Species Extinction. Chapman and Hall,London, pp. 119142.
Buckland, S.T., Anderson, D.R., Burnham, K.P., Laake, J.L., 1993.Distance Sampling, Estimating Abundance of Biological Popula-tions Chapman and Hall, London.
Chiarello, A.G., 1994. Diet of the brown howler monkey Alouattafusca in a semi-deciduous forest fragment of southeastern Brazil.Primates 35, 2534.
Chiarello, A.G., 1997. Mammalian community and forest structure ofAtlantic forest fragments in south-eastern Brazil. Ph.D. dissertation,University of Cambridge, Cambridge, UK.
Chiarello, A.G., 1998. Diet of the Atlantic forest maned sloth, Brady-pus torquatus (Xenarthra: Bradypodidae). Journal of Zoology 246,1119.
Chiarello, A.G., Passamani, M., Zorte a, M., 1997. Field observationson the thin-spined porcupine, Chaetomys subspinosus (Rodentia;Echimyidae). Mammalia 61, 2936.
Dean, W., 1995. With the Broadax and Firebrand, the Destruction ofthe Brazilian Atlantic Forest. University of California Press, Berke-ley.
Dillon, W.R., Goldstein, M., 1984. Multivariate Analysis: Methodsand Applications. John Wiley and Sons, New York.
Dirzo, R., Miranda, A., 1990. Comtemporary Neotropical defauna-tion and forest structure, function and diversitya sequel to JohnTerborgh. Conservation Biology 4, 444447.
Eisenberg, J.F., 1981. The Mammalian Radiations: An Analysis ofTrends in Evolution, Adaptation, and Behavior. The University ofChicago, London.
Emmons, L.H., Feer, F., 1997. Neotropical Rainforest Mammals, AField Guide, 2nd ed. The University of Chicago Press, Chicago.
Fonseca, G.A.B., 1985. The vanishing Brazilian Atlantic forest.Biological Conservation 34, 1734.
Fonseca, G.A.B., Robinson, J.G., 1990. Forest size and structure:competitive and predatory eects on small mammal communities.Biological Conservation 53, 265294.
Frankham, R., 1998. Inbreeding and extinction: island populations.Conservation Biology 12, 665675.
Franklin, I.R., Frankham, R., 1998. How large must populations
be to retain evolutionary potential? Animal Conservation 1, 6973.Gentry, A.H., Emmons, L.H., 1987. Geographical variation in ferti-
lity, phenology, and composition of the understory of Neotropicalforests. Biotropica 19, 216227.
Glanz, W.E., 1990. Neotropical mammal densities: how unusual is thecommunity on Barro Colorado Island, Panama? In: Gentry, 1990(Ed.), Four Neotropical Rainforests. Yale University Press, NewHaven, pp. 287313.
Hallwachs, W., 1986. Agoutis (Dasyprocta punctata): the inheritors ofguapinol (Hymenaea courbaril: Leguminosae). In: Estrada, A.,Fleming, T.H. (Eds.), Frugivores and Seed Dispersal. Dr. W. Junk,Dordredcht, The Netherlands, pp. 285304.
Harris, L.D., 1984. The Fragmented Forest, Island BiogeographyTheory and the Preservation of Biotic Diversity. The University ofChicago Press, Chicago.
Heinsdijk, D., de Macedo, J.G., Andel, S., Ascoly, R.B., 1965. A
Floresta do Norte do Espirito Santo. Departamento de RecursosNaturais Renova veisDivisao de Silvicultura, Sec ao de PesquisasFlorestais, Rio de Janeiro.
Janson, C.H., Emmons, L.H., 1990. Ecological structure of the non-ying mammal community at Cocha Cashu, Peru. In: Gentry, A.H.(Ed.), Four Neotropical Rainforests. Yale University Press, NewHaven, pp. 314338.
Kent, M., Coker, P., 1996. Vegetation Description and Analysis, aPractical Approach. John Wiley and Sons, Chichester.
Laurance, W.F., 1990. Comparative responses of ve arboreal marsu-pials to tropical forest fragmentation. Journal of Mammalogy 71,641653.
Lovejoy, T.E., Bierregaard, R.O., Rylands, A.B., Malcolm, J.R.,Quintela, C.E., Harper, L.H., Brown Jr, K.S., Powell, A.H., Powell,G.V.N., Schubart, H.O.R., Hays, M.B., 1986. Edge and othereects of isolation on Amazon forest fragments. In: Soule , M.E.(Ed.), Conservation Biology, the Science of Scarcity and Diversity.Sinauer Associates, Sunderland, MA, pp. 257285.
MacArthur, R.H., Wilson, E.O., 1967. The Theory of Island Biogeo-graphy Princeton University Press, Princeton, N.J.
MacArthur, R.H., Diamond, J.M., Karr, J.R., 1972. Density com-pensation in island faunas. Ecology 53, 330342.
Malcolm, J.R., 1988. Small mammal abundance in isolated and non-isolated primary forest reserves near Manaus, Brazil. Acta Amazonica18, 6783.
Matthiae, P.E., Stearns, F., 1981. Mammals in forest islands insouthestern Wisconsin. In: Burgess, R.L., Sharpe, D.M. (Eds.),Forest Island Dynamics in Man-Dominated Landscapes. Springer-Verlag, New York, pp. 5566.
McNab, B.K., 1963. Bioenergetics and the determination of homerange size. American Naturalist 97, 133140.
Milton, K., May, M., 1976. Body weight, diet and home range area inprimates. Nature 259, 459462.
Mittermeier, R.A., Myers, N., Thomsen, J.B., Fonseca, G.A.B., Oli-vieri, S., 1998. Biodiversity hotspots and major tropical wildernessareas: approaches to setting conservation priorities. ConservationBiology 12, 516520.
Moraes, C., 1974. Geograa do Esprito Santo, 1st ed. Fundac aoCultural do Esprito Santo, Vito ria, Brazil.
Norusis, M., 1993. SPSS for Windows, Base System User's Guide,Release 6.0. Marketing Department SPSS, Chicago, IL.
Oliver, W.L.R., Santos, I.B., 1991. Threatened endemic mammals ofthe Atlantic forest region of south-eastern Brazil. Wildlife Pre-servation Trust Special Scientic Report 4, 1126.
Paine, R.T., 1966. Food web complexity and species diversity. Amer-ican Naturalist 100, 6575.
Patterson, B.D., 1987. The principle of nested subsets and its implica-tions for biological conservation. Conservation Biology 1, 323334.
A.G. Chiarello/Biological Conservation 89 (1999) 7182 81
8/3/2019 1999E ects of fragmentation of the Atlantic forest on mammal
12/12
Putz, F.E., Leigh, E.G.Jr., Wright, S.J., 1990. Solitary connement inPanama. Garden 14, 1823.
Redford, K.H., 1992. The empty forest. Bioscience 42, 412422.Redford, K.H., Robinson, J.G., 1987. The game of choice: patterns of
indian and colonist hunting in the Neotropics. American Anthro-pologist 89, 650667.
Redford, K.H., Robinson, J.G., 1991. Park size and the conservationof forest mammals in Latin America. In: Mares, M.A., Schmidly,
D.J. (Eds.), Latin American Mammalogy, History, Biodiversity,and Conervation. University of Oklahoma Press, Norman, OK andLondon, pp. 227234.
Rizzini, C.T., 1963. Nota pre via sobre a divisao togeogra ca (or-sticasociolo gica) do Brasil. Revista Brasileira de Geograa 25, 364.
Robinson, J.G., 1996. Hunting wildlife in forest patches: an ephemeralresource. In: Schelhas, J., Greenberg, R. (Eds.), Forest Patches inTropical Landscapes. Island Press, Washington, DC, pp. 111130.
Robinson, J.G., Redford, K.H., 1986. Body size, diet, and populationdensity of Neotropical forest mammals. American Naturalist 128,665680.
Ruschi, A., 1978. A atual fauna de mamferos, aves e re pteis dareserva biolo gica de Comboios. Boletim do Museu de BiologiaMello Leitao, Se rie Zoologia 90, 126.
Saint-Hilaire, A.F.C., 1833. Voyage dans le district des diamans et surle littoral du Bre sil, VIIXV, H. Herluison (Ed.), Paris.
SOS Mata Atla ntica and INPE, 1997. Remanescentes de Mata Atla n-tica e Ecossistemas Associados-Perodo 19901995. Map. Fundac aoSOS Mata Atla ntica and Instituto Nacional de Pesquisas Espaciais,Sao Paulo, Brazil.
SOS Mata Atla ntica, INPE and IS, 1998. Atlas da Evoluc ao dosRemanescentes Florestais e Ecossistemas Associados do Domnioda Mata Atla ntica no Perodo 19901995. Fundac ao SOS MataAtla ntica, Instituto Nacional de Pesquisas Espaciais e Instituto
Socioambiental, Sao Paulo, Brazil.Terborgh, J., Winter, B., 1980. Some causes of extinction. In: Soule ,M.E., Wilcox, B.A. (Eds.), Conservation Biology, an Evolutionary-Ecological Perspective. Sinauer, Sunderland, pp. 119133.
Wied-Neuwied, P.M., 1820. Travels in Brazil in 1815, 1816, and 1817,[Translated from the German], Sir Richard Phillips and Company,London.
Willis, E.O., 1979. The composition of avian communities in rema-nescent woodlots in southern Brazil. Pape is Avulsos de Zoologia,Sao Paulo 33, 125.
Wilson, D.E., Reeder, D.A.M. (Eds.), 1993. Mammal Species of theWorld: a Taxonomic andGeographic Reference, 2nded., SmithsonianInstitution Press, Washington, DC.
Wilson, E.O., Willis, E.O., 1975. Applied Biogeography. In: Cody,M.L., Diamond, J.M. (Eds.), Ecology and Evolution of Commu-nities. Harvard University Press, Cambridge, USA, pp. 522534.
82 A.G. Chiarello/Biological Conservation 89 (1999) 7182
Top Related